]> git.ipfire.org Git - thirdparty/pdns.git/blob - pdns/dnsdist-carbon.cc
dnsdist: Add metrics about TLS handshake failures for DoH and DoT
[thirdparty/pdns.git] / pdns / dnsdist-carbon.cc
1 /*
2 * This file is part of PowerDNS or dnsdist.
3 * Copyright -- PowerDNS.COM B.V. and its contributors
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of version 2 of the GNU General Public License as
7 * published by the Free Software Foundation.
8 *
9 * In addition, for the avoidance of any doubt, permission is granted to
10 * link this program with OpenSSL and to (re)distribute the binaries
11 * produced as the result of such linking.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
21 */
22 #ifdef HAVE_CONFIG_H
23 #include "config.h"
24 #endif
25 #include "iputils.hh"
26 #include "dolog.hh"
27 #include "sstuff.hh"
28
29 #include "namespaces.hh"
30 #include "dnsdist.hh"
31 #include "threadname.hh"
32
33 GlobalStateHolder<vector<CarbonConfig> > g_carbon;
34 static time_t s_start=time(0);
35 uint64_t uptimeOfProcess(const std::string& str)
36 {
37 return time(0) - s_start;
38 }
39
40 void carbonDumpThread()
41 try
42 {
43 setThreadName("dnsdist/carbon");
44 auto localCarbon = g_carbon.getLocal();
45 for(int numloops=0;;++numloops) {
46 if(localCarbon->empty()) {
47 sleep(1);
48 continue;
49 }
50 /* this is wrong, we use the interval of the first server
51 for every single one of them */
52 if(numloops) {
53 const unsigned int interval = localCarbon->at(0).interval;
54 sleep(interval);
55 }
56
57 for (const auto& conf : *localCarbon) {
58 const auto& server = conf.server;
59 const std::string& namespace_name = conf.namespace_name;
60 std::string hostname = conf.ourname;
61 if(hostname.empty()) {
62 char tmp[80];
63 memset(tmp, 0, sizeof(tmp));
64 gethostname(tmp, sizeof(tmp));
65 char *p = strchr(tmp, '.');
66 if(p) *p=0;
67 hostname=tmp;
68 boost::replace_all(hostname, ".", "_");
69 }
70 const std::string& instance_name = conf.instance_name;
71
72 try {
73 Socket s(server.sin4.sin_family, SOCK_STREAM);
74 s.setNonBlocking();
75 s.connect(server); // we do the connect so the attempt happens while we gather stats
76 ostringstream str;
77 time_t now=time(0);
78 for(const auto& e : g_stats.entries) {
79 str<<namespace_name<<"."<<hostname<<"."<<instance_name<<"."<<e.first<<' ';
80 if(const auto& val = boost::get<DNSDistStats::stat_t*>(&e.second))
81 str<<(*val)->load();
82 else if (const auto& dval = boost::get<double*>(&e.second))
83 str<<**dval;
84 else
85 str<<(*boost::get<DNSDistStats::statfunction_t>(&e.second))(e.first);
86 str<<' '<<now<<"\r\n";
87 }
88 auto states = g_dstates.getLocal();
89 for(const auto& state : *states) {
90 string serverName = state->name.empty() ? (state->remote.toString() + ":" + std::to_string(state->remote.getPort())) : state->getName();
91 boost::replace_all(serverName, ".", "_");
92 const string base = namespace_name + "." + hostname + "." + instance_name + ".servers." + serverName + ".";
93 str<<base<<"queries" << ' ' << state->queries.load() << " " << now << "\r\n";
94 str<<base<<"responses" << ' ' << state->responses.load() << " " << now << "\r\n";
95 str<<base<<"drops" << ' ' << state->reuseds.load() << " " << now << "\r\n";
96 str<<base<<"latency" << ' ' << (state->availability != DownstreamState::Availability::Down ? state->latencyUsec/1000.0 : 0) << " " << now << "\r\n";
97 str<<base<<"senderrors" << ' ' << state->sendErrors.load() << " " << now << "\r\n";
98 str<<base<<"outstanding" << ' ' << state->outstanding.load() << " " << now << "\r\n";
99 str<<base<<"tcpdiedsendingquery" << ' '<< state->tcpDiedSendingQuery.load() << " " << now << "\r\n";
100 str<<base<<"tcpdiedreaddingresponse" << ' '<< state->tcpDiedReadingResponse.load() << " " << now << "\r\n";
101 str<<base<<"tcpgaveup" << ' '<< state->tcpGaveUp.load() << " " << now << "\r\n";
102 str<<base<<"tcpreadimeouts" << ' '<< state->tcpReadTimeouts.load() << " " << now << "\r\n";
103 str<<base<<"tcpwritetimeouts" << ' '<< state->tcpWriteTimeouts.load() << " " << now << "\r\n";
104 str<<base<<"tcpcurrentconnections" << ' '<< state->tcpCurrentConnections.load() << " " << now << "\r\n";
105 str<<base<<"tcpavgqueriesperconnection" << ' '<< state->tcpAvgQueriesPerConnection.load() << " " << now << "\r\n";
106 str<<base<<"tcpavgconnectionduration" << ' '<< state->tcpAvgConnectionDuration.load() << " " << now << "\r\n";
107 }
108
109 std::map<std::string,uint64_t> frontendDuplicates;
110 for(const auto& front : g_frontends) {
111 if (front->udpFD == -1 && front->tcpFD == -1)
112 continue;
113
114 string frontName = front->local.toString() + ":" + std::to_string(front->local.getPort()) + (front->udpFD >= 0 ? "_udp" : "_tcp");
115 boost::replace_all(frontName, ".", "_");
116 auto dupPair = frontendDuplicates.insert({frontName, 1});
117 if (!dupPair.second) {
118 frontName = frontName + "_" + std::to_string(dupPair.first->second);
119 ++(dupPair.first->second);
120 }
121
122 const string base = namespace_name + "." + hostname + "." + instance_name + ".frontends." + frontName + ".";
123 str<<base<<"queries" << ' ' << front->queries.load() << " " << now << "\r\n";
124 str<<base<<"responses" << ' ' << front->responses.load() << " " << now << "\r\n";
125 str<<base<<"tcpdiedreadingquery" << ' '<< front->tcpDiedReadingQuery.load() << " " << now << "\r\n";
126 str<<base<<"tcpdiedsendingresponse" << ' '<< front->tcpDiedSendingResponse.load() << " " << now << "\r\n";
127 str<<base<<"tcpgaveup" << ' '<< front->tcpGaveUp.load() << " " << now << "\r\n";
128 str<<base<<"tcpclientimeouts" << ' '<< front->tcpClientTimeouts.load() << " " << now << "\r\n";
129 str<<base<<"tcpdownstreamtimeouts" << ' '<< front->tcpDownstreamTimeouts.load() << " " << now << "\r\n";
130 str<<base<<"tcpcurrentconnections" << ' '<< front->tcpCurrentConnections.load() << " " << now << "\r\n";
131 str<<base<<"tcpavgqueriesperconnection" << ' '<< front->tcpAvgQueriesPerConnection.load() << " " << now << "\r\n";
132 str<<base<<"tcpavgconnectionduration" << ' '<< front->tcpAvgConnectionDuration.load() << " " << now << "\r\n";
133 str<<base<<"tls10-queries" << ' ' << front->tls10queries.load() << " " << now << "\r\n";
134 str<<base<<"tls11-queries" << ' ' << front->tls11queries.load() << " " << now << "\r\n";
135 str<<base<<"tls12-queries" << ' ' << front->tls12queries.load() << " " << now << "\r\n";
136 str<<base<<"tls13-queries" << ' ' << front->tls13queries.load() << " " << now << "\r\n";
137 str<<base<<"tls-unknown-queries" << ' ' << front->tlsUnknownqueries.load() << " " << now << "\r\n";
138 str<<base<<"tlsnewsessions" << ' ' << front->tlsNewSessions.load() << " " << now << "\r\n";
139 str<<base<<"tlsresumptions" << ' ' << front->tlsResumptions.load() << " " << now << "\r\n";
140 str<<base<<"tlsunknownticketkeys" << ' ' << front->tlsUnknownTicketKey.load() << " " << now << "\r\n";
141 str<<base<<"tlsinactiveticketkeys" << ' ' << front->tlsInactiveTicketKey.load() << " " << now << "\r\n";
142 const TLSErrorCounters* errorCounters = nullptr;
143 if (front->tlsFrontend != nullptr) {
144 errorCounters = &front->tlsFrontend->d_tlsCounters;
145 }
146 else if (front->dohFrontend != nullptr) {
147 errorCounters = &front->dohFrontend->d_tlsCounters;
148 }
149 if (errorCounters != nullptr) {
150 str<<base<<"tlsdhkeytoosmall" << ' ' << errorCounters->d_dhKeyTooSmall << " " << now << "\r\n";
151 str<<base<<"tlsinappropriatefallback" << ' ' << errorCounters->d_inappropriateFallBack << " " << now << "\r\n";
152 str<<base<<"tlsnosharedcipher" << ' ' << errorCounters->d_noSharedCipher << " " << now << "\r\n";
153 str<<base<<"tlsunknownciphertype" << ' ' << errorCounters->d_unknownCipherType << " " << now << "\r\n";
154 str<<base<<"tlsunknownkeyexchangetype" << ' ' << errorCounters->d_unknownKeyExchangeType << " " << now << "\r\n";
155 str<<base<<"tlsunknownprotocol" << ' ' << errorCounters->d_unknownProtocol << " " << now << "\r\n";
156 str<<base<<"tlsunsupportedec" << ' ' << errorCounters->d_unsupportedEC << " " << now << "\r\n";
157 str<<base<<"tlsunsupportedprotocol" << ' ' << errorCounters->d_unsupportedProtocol << " " << now << "\r\n";
158 }
159 }
160
161 auto localPools = g_pools.getLocal();
162 for (const auto& entry : *localPools) {
163 string poolName = entry.first;
164 boost::replace_all(poolName, ".", "_");
165 if (poolName.empty()) {
166 poolName = "_default_";
167 }
168 const string base = namespace_name + "." + hostname + "." + instance_name + ".pools." + poolName + ".";
169 const std::shared_ptr<ServerPool> pool = entry.second;
170 str<<base<<"servers" << " " << pool->countServers(false) << " " << now << "\r\n";
171 str<<base<<"servers-up" << " " << pool->countServers(true) << " " << now << "\r\n";
172 if (pool->packetCache != nullptr) {
173 const auto& cache = pool->packetCache;
174 str<<base<<"cache-size" << " " << cache->getMaxEntries() << " " << now << "\r\n";
175 str<<base<<"cache-entries" << " " << cache->getEntriesCount() << " " << now << "\r\n";
176 str<<base<<"cache-hits" << " " << cache->getHits() << " " << now << "\r\n";
177 str<<base<<"cache-misses" << " " << cache->getMisses() << " " << now << "\r\n";
178 str<<base<<"cache-deferred-inserts" << " " << cache->getDeferredInserts() << " " << now << "\r\n";
179 str<<base<<"cache-deferred-lookups" << " " << cache->getDeferredLookups() << " " << now << "\r\n";
180 str<<base<<"cache-lookup-collisions" << " " << cache->getLookupCollisions() << " " << now << "\r\n";
181 str<<base<<"cache-insert-collisions" << " " << cache->getInsertCollisions() << " " << now << "\r\n";
182 str<<base<<"cache-ttl-too-shorts" << " " << cache->getTTLTooShorts() << " " << now << "\r\n";
183 }
184 }
185
186 #ifdef HAVE_DNS_OVER_HTTPS
187 {
188 std::map<std::string,uint64_t> dohFrontendDuplicates;
189 const string base = "dnsdist." + hostname + ".main.doh.";
190 for(const auto& doh : g_dohlocals) {
191 string name = doh->d_local.toStringWithPort();
192 boost::replace_all(name, ".", "_");
193 boost::replace_all(name, ":", "_");
194 boost::replace_all(name, "[", "_");
195 boost::replace_all(name, "]", "_");
196
197 auto dupPair = dohFrontendDuplicates.insert({name, 1});
198 if (!dupPair.second) {
199 name = name + "_" + std::to_string(dupPair.first->second);
200 ++(dupPair.first->second);
201 }
202
203 vector<pair<const char*, const std::atomic<uint64_t>&>> v{
204 {"http-connects", doh->d_httpconnects},
205 {"http1-queries", doh->d_http1Stats.d_nbQueries},
206 {"http2-queries", doh->d_http2Stats.d_nbQueries},
207 {"http1-200-responses", doh->d_http1Stats.d_nb200Responses},
208 {"http2-200-responses", doh->d_http2Stats.d_nb200Responses},
209 {"http1-400-responses", doh->d_http1Stats.d_nb400Responses},
210 {"http2-400-responses", doh->d_http2Stats.d_nb400Responses},
211 {"http1-403-responses", doh->d_http1Stats.d_nb403Responses},
212 {"http2-403-responses", doh->d_http2Stats.d_nb403Responses},
213 {"http1-500-responses", doh->d_http1Stats.d_nb500Responses},
214 {"http2-500-responses", doh->d_http2Stats.d_nb500Responses},
215 {"http1-502-responses", doh->d_http1Stats.d_nb502Responses},
216 {"http2-502-responses", doh->d_http2Stats.d_nb502Responses},
217 {"http1-other-responses", doh->d_http1Stats.d_nbOtherResponses},
218 {"http2-other-responses", doh->d_http2Stats.d_nbOtherResponses},
219 {"get-queries", doh->d_getqueries},
220 {"post-queries", doh->d_postqueries},
221 {"bad-requests", doh->d_badrequests},
222 {"error-responses", doh->d_errorresponses},
223 {"redirect-responses", doh->d_redirectresponses},
224 {"valid-responses", doh->d_validresponses}
225 };
226
227 for(const auto& item : v) {
228 str<<base<<name<<"."<<item.first << " " << item.second << " " << now <<"\r\n";
229 }
230 }
231 }
232 #endif /* HAVE_DNS_OVER_HTTPS */
233
234 {
235 WriteLock wl(&g_qcount.queryLock);
236 std::string qname;
237 for(auto &record: g_qcount.records) {
238 qname = record.first;
239 boost::replace_all(qname, ".", "_");
240 str<<"dnsdist.querycount." << qname << ".queries " << record.second << " " << now << "\r\n";
241 }
242 g_qcount.records.clear();
243 }
244
245 const string msg = str.str();
246
247 int ret = waitForRWData(s.getHandle(), false, 1 , 0);
248 if(ret <= 0 ) {
249 vinfolog("Unable to write data to carbon server on %s: %s", server.toStringWithPort(), (ret<0 ? stringerror() : "Timeout"));
250 continue;
251 }
252 s.setBlocking();
253 writen2(s.getHandle(), msg.c_str(), msg.size());
254 }
255 catch(std::exception& e) {
256 warnlog("Problem sending carbon data: %s", e.what());
257 }
258 }
259 }
260 }
261 catch(std::exception& e)
262 {
263 errlog("Carbon thread died: %s", e.what());
264 }
265 catch(PDNSException& e)
266 {
267 errlog("Carbon thread died, PDNSException: %s", e.reason);
268 }
269 catch(...)
270 {
271 errlog("Carbon thread died");
272 }